Update Assembler::TryAllocate to support inline allocation tracing

Update Assembler::TryAllocate to support inline allocation tracing

- Track whether class allocation should be traced in existing ClassHeapStats structure.
- Emit compare and branch at top of ::TryAllocate

These changes appear to be performance neutral.

An alternative approach is to deoptimize all code that inlined allocations for a specific class. I do not think this is a good approach because we will spoil the water- allocations that aren't occurring in the (optimized steady state) running program  will appear in the allocation profile. This will make the data much less actionable bordering on misleading. The approach used in this CL will guarantee the allocation profile aligns with the actual running program.

R=regis@google.com, srdjan@google.com

Committed: https://github.com/dart-lang/sdk/commit/598c67001b5154fca65108d52352c03b6e0fad97

[Cutch, 2015-07-06 17:07:09]

johnmccutchan@google.com changed reviewers:
+ srdjan@google.com

[Cutch, 2015-07-06 20:25:10]

johnmccutchan@google.com changed reviewers:
+ fschneider@google.com

[Cutch, 2015-07-06 21:43:44]

I've done this for IA32 only. Once I get an LGTM I will port to other
architectures.

[srdjan, 2015-07-06 22:23:45]

LGTM

https://codereview.chromium.org/1213013002/diff/80001/runtime/vm/assembler_ia...
File runtime/vm/assembler_ia32.h (right):

https://codereview.chromium.org/1213013002/diff/80001/runtime/vm/assembler_ia...
runtime/vm/assembler_ia32.h:877: bool near_jump);
Document when is 'failure' jumped to, I would give the label a different name
than failure.

[Florian Schneider, 2015-07-07 12:43:21]

dbc:

https://codereview.chromium.org/1213013002/diff/80001/runtime/vm/assembler_ia...
File runtime/vm/assembler_ia32.cc (right):

https://codereview.chromium.org/1213013002/diff/80001/runtime/vm/assembler_ia...
runtime/vm/assembler_ia32.cc:2718: testb(state_address,
Immediate(ClassHeapStats::TraceAllocationMask()));
state_ is a intptr_t, and you're using testb. Even though it works for this
architecture, I'd make state_ a 1-byte size type bool or char for consistency.

https://codereview.chromium.org/1213013002/diff/80001/runtime/vm/class_table.h
File runtime/vm/class_table.h (right):

https://codereview.chromium.org/1213013002/diff/80001/runtime/vm/class_table....
runtime/vm/class_table.h:140: intptr_t state_;
bool or char?

[Cutch, 2015-07-09 23:37:34]

https://codereview.chromium.org/1213013002/diff/80001/runtime/vm/assembler_ia...
File runtime/vm/assembler_ia32.cc (right):

https://codereview.chromium.org/1213013002/diff/80001/runtime/vm/assembler_ia...
runtime/vm/assembler_ia32.cc:2718: testb(state_address,
Immediate(ClassHeapStats::TraceAllocationMask()));
On 2015/07/07 12:43:21, Florian Schneider wrote:
> state_ is a intptr_t, and you're using testb. Even though it works for this
> architecture, I'd make state_ a 1-byte size type bool or char for consistency.

I'd rather not change the size of the state_ field. I'm only using testb on IA32
and X64.

https://codereview.chromium.org/1213013002/diff/80001/runtime/vm/assembler_ia...
File runtime/vm/assembler_ia32.h (right):

https://codereview.chromium.org/1213013002/diff/80001/runtime/vm/assembler_ia...
runtime/vm/assembler_ia32.h:877: bool near_jump);
On 2015/07/06 22:23:45, srdjan wrote:
> Document when is 'failure' jumped to, I would give the label a different name
> than failure.

Done.

[Cutch, 2015-07-09 23:37:48]

Ported to other architectures. PTAL.

[Cutch, 2015-07-10 17:13:46]

johnmccutchan@google.com changed reviewers:
+ regis@google.com

[Cutch, 2015-07-10 17:13:55]

+regis for ARM, ARM64, and MIPS

[srdjan, 2015-07-10 17:31:22]

lgtm

https://codereview.chromium.org/1213013002/diff/100001/runtime/vm/class_table.cc
File runtime/vm/class_table.cc (right):

https://codereview.chromium.org/1213013002/diff/100001/runtime/vm/class_table...
runtime/vm/class_table.cc:242: state_ = 0;
Could you please add here also: 
old_pre_new_gc_count_ = 0;
old_pre_new_gc_size_ = 0;

[regis, 2015-07-10 18:15:26]

lgtm

[Cutch, 2015-07-10 19:19:06]

Committed patchset #7 (id:120001) manually as
598c67001b5154fca65108d52352c03b6e0fad97 (presubmit successful).

[Ivan Posva, 2015-07-16 11:34:47]

iposva@google.com changed reviewers:
+ iposva@google.com

[Ivan Posva, 2015-07-16 11:34:47]

DBC -ip

https://codereview.chromium.org/1213013002/diff/80001/runtime/vm/object.cc
File runtime/vm/object.cc (right):

https://codereview.chromium.org/1213013002/diff/80001/runtime/vm/object.cc#ne...
runtime/vm/object.cc:2737: TraceAllocationBit::update(trace_allocation,
raw_ptr()->state_bits_));
Why is this not under "if (changed)"?

[Cutch, 2015-07-16 14:34:07]

https://codereview.chromium.org/1213013002/diff/80001/runtime/vm/object.cc
File runtime/vm/object.cc (right):

https://codereview.chromium.org/1213013002/diff/80001/runtime/vm/object.cc#ne...
runtime/vm/object.cc:2737: TraceAllocationBit::update(trace_allocation,
raw_ptr()->state_bits_));
On 2015/07/16 11:34:47, Ivan Posva wrote:
> Why is this not under "if (changed)"?

My bad- https://codereview.chromium.org/1234943004/